Research in the Unit of Neural Network Physiology is primarily concerned with the cortex-basal ganglia system, which is important for movement control, reward mediated behavior, and higher cognitive functions. Basal ganglia dysfunctions, e.g. those that arise from a dopamine imbalance in the striatum, are correlated with severe movement disorders, e.g. Parkinson's disease and cognitive disorders e.g. Schizophrenia. Despite their involvement in a variety of higher motor and cognitive functions, the functional architecture of the cortex - basal ganglia system is highly parallel suggesting a common principle in information processing. This principle is reflected in several parallel cognitive and motor loops in which the basal ganglia receive inputs from the cortex, and basal ganglia outputs through thalamus control cortical activity. In our work, we reconstruct parts of these loops in vitro by culturing young rat or mouse brains for up to several months. These cultures provide the most complex in vitro system that exists to date: a 6-layered cortical culture that drives activity in a striatal culture and also receives dopaminergic inputs from the substantia nigra. The system comprises of several hundred thousand of neurons and replicates network activity that strongly resembles that seen in vivo. Taking advantage of this in vitro approach, we are in the unique position to study single neuron electrophysiology, synaptic transmission between neurons, and neuronal populations within and across nuclei under in vivo-like conditions. This year's research has focused primarily on two aspects of striatal function. The striatum is the first stage in basal ganglia circuits, which processes cortical inputs. Therefore, understanding the factors that control striatal responses to cortical inputs is crucial for our understanding of basal ganglia function. Until recently, it was assumed that cortical inputs depolarize neurons above firing threshold and the resulting action potential immediately leaves the striatum in order to inhibit neurons in basal ganglia output nuclei. Our work has shown that this view of striatal function is too simplistic and that striatal action potentials contribute significantly to local striatal processing. First, we demonstrated that the action potential propagates from the soma not only down the axon, but also up into the dendrite. We showed that the intracellular calcium concentration in higher order dendrites encodes the exact number of action potentials at the soma. Through this mechanism, the dendrites are informed about spike output at the soma. Because cortical inputs to the striatum are highly plastic and spike backpropagation has been demonstrated to play an important role in regulating such plasticity in the cortex, our findings for the first time that suggest spike backpropagation might also apply for the regulation of corticostriatal plasticity at the striatal level. Second, striatal neurons possess extensive local axon collaterals, but their function has been elusive. Because these collaterals are extremely important with respect to neural network theories on striatal function, several research groups tried to characterize synaptic transmission between these neurons, but failed. We have now characterized in acute slices and cortex-striatum-substantia nigra triple cultures the synaptic nature of these collaterals and demonstrated that striatal neurons control firing between local neighbors through a fast GABAergic synaptic transmission. Because an imbalance of striatal GABAergic activity is at the core of many basal ganglia diseases, the demonstration of this synaptic connections has widespread implications on striatal function and dysfunction.